Decadal-scale responses of larval fish assemblages to multiple ...

MARINE ECOLOGY PROGRESS SERIES Mar Ecol Prog Ser

Vol. 450: 37–53, 2012 doi: 10.3354/meps09540

Published March 29

Decadal-scale responses of larval fish assemblages to multiple ecosystem processes in the northern Gulf of Mexico Barbara A. Muhling1,*, John T. Lamkin2, William J. Richards2 1

Cooperative Institute for Marine and Atmospheric Studies, University of Miami, Miami, Florida 33149, USA 2 NOAA/NMFS/Southeast Fisheries Science Center, Miami, Florida 33149, USA

ABSTRACT: Larval fish assemblages have been systematically sampled across the continental shelf of the northern Gulf of Mexico since the early 1980s. To date, these data have not been analyzed in terms of assemblage structure or variability, despite representing a highly valuable resource for examining decadal-scale change. In this study, multivariate statistical techniques were used to characterize temporal and spatial changes in abundances of the larvae of 20 common fish families from the early 1980s through to the late 2000s. The larvae of some pelagic and mesopelagic families showed marked increases in abundance over the survey time period, while the abundances of some benthic families decreased. Changes in assemblage structure were partially explained by changes in the Gulf of Mexico environment with respect to sea surface temperature and changes in shrimp trawling effort. Outflow from the Mississippi River was also influential on the interannual assemblage variability. However, the strong directional trends apparent in many family groups remained unexplained, and further research is required to discern the drivers of these patterns. KEY WORDS: Gulf of Mexico · Ichthyoplankton · Multivariate statistics · Community analyses Resale or republication not permitted without written consent of the publisher

Marine ecosystems around the world are subject to complex environmental variability, characterized by changes in sea temperature, ocean currents, species composition and food web structure (Francis & Hare 1994, Anderson & Piatt 1999, Beaugrand 2004). As a result, marine populations frequently fluctuate in size through time, as conditions for larval survival and recruitment vary, and adults respond to changes in habitat suitability (Smith & Moser 2003). Complicating these signals are additional pressures from extractive fisheries. As economically valuable species are preferentially removed, previous interactions with predator, prey and competitor species are altered, leading to cascading effects throughout food webs (Essington et al. 2002, Hinke et al. 2004). In addition, directional processes, such as

anthropogenic climate change, affect species assemblages and interactions, as species tolerant of warmer waters slowly displace those with cooler preferences (Nye et al. 2009, Fodrie et al. 2010). These mechanisms are frequently species-specific and involve many interactions and feedback responses, resulting in complex, non-linear trends of species assemblages through time (Levin et al. 2006). As a result, although abundance indices of commercially important species or trends in easily sampled environmental variables provide some information on ecosystem state, it has been historically difficult to formulate useful and encompassing ecosystem indicators (Hilty & Merenlender 2000). Most marine fishes, regardless of eventual adult habitat, have a pelagic larval phase. Multi-year time series of larval fish abundances may track temporal changes in adult biomass across a wide variety of fish

*Email: [email protected]

© Inter-Research 2012 · www.int-res.com

INTRODUCTION

38

Mar Ecol Prog Ser 450: 37–53, 2012

functional groups, including exploited and unexMETHODS ploited species from a wide range of adult habitats (Smith & Moser 2003). Larval fish assemblages thus Data collection have great potential for use as response indicators to changes in environmental conditions and changes in Plankton samples have been collected across the fishing pressure. northern Gulf of Mexico under the Southeast Area In this study, spatial and temporal trends in larval Monitoring and Assessment Program (SEAMAP) fish assemblages sampled on the northern Gulf of since 1982 (Lyczkowski-Shultz & Hanisko 2007; Mexico continental shelf were examined for eviFig. 1). Sampling effort has largely concentrated on dence of decadal-scale shifts. The Gulf of Mexico is a the spring (April to May), early summer (June to large, semi-enclosed basin, which is connected to the July), late summer (August to September) and fall Caribbean Sea in the south and the Atlantic Ocean to (October to November) time periods (Table 1). The the east via the Loop Current. It supports many complankton sampling was completed using bongo nets, mercial and recreational fisheries, most notably shrimp trawling on the continental shelf and pelagic longlining for tuna and swordfish further offshore (Alverson et al. 1996). The continental shelf environment is dominated by seasonal changes in water temperature, as a result of surface heating, and seasonal discharge of low salinity water from rivers onto the northern and eastern shelf (Müller-Karger et al. 1991). Nutrients from this inflow have contributed to the formation of a large hypoxic zone on the Texas-Louisiana shelf, which has expanded in size since the 1980s (Rabalais et al. 2002). In addition, recent studies of inshore species and communities have suggested decadal-scale shifts as a result of warming water temperatures (Cuevas et al. 2004, Fodrie et al. 2010). However, spatial and temporal changes in the broader fish communities of the northern Gulf of Mexico remain largely unknown. The objective of this study was to characterize spatial and temporal changes in larval fish assemblages on the northern Gulf of Mexico continental shelf from the early 1980s through the late 2000s. We aimed to relate trends in assemblages to environmental metrics and to formulate multivariate ecosystem indicators from larval fish data. It was hypothesized that larvae of fish families occupying similar habitats as adults would show similar decadal-scale changes across the Fig. 1. Sampling effort in the Gulf of Mexico SEAMAP program, 1984 to 2008. time series and that larvae of benthic (a) Total plankton samples collected, rounded to nearest 0.1°. The solid confamilies would decrease through time, in tour line is 2000 m depth; the dashed contour is 64 m depth (boundary of innershelf and outer-shelf zones). (b) Mean sampled depth and (c) mean sampled response to declining habitat quality date of outer-shelf stations occupied during May cruises, to show interannual caused by bottom hypoxia. variability in sampling effort. Samples were collected in all years except 1985

Muhling et al.: Gulf of Mexico ichthyoplankton assemblages

39

Table 1. Number of samples collected by year and month through SEAMAP surveys from 1984 to 2008. Five time series with consistent interannual coverage were identified and are shown in gray—ESu ISW: Early Summer Inner Shelf (West); Su IS: Summer Inner Shelf; Fall ISW: Fall Inner Shelf (West); Sp OS: Spring Outer Shelf; Su OS: Summer Outer Shelf. Notes indicate sample subset included in the respective time series Inner shelf (Shallower than 64 m depth) Apr May Jun Jul Aug Sep Oct Nov Dec 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2002 2003 2004 2005 2006 2007 2008

0 0 0 0 0 0 4 4 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0

Sufficient No east/west coverage? Time series name Notes

Apr

Outer shelf (Deeper than 64 m depth) May Jun Jul Aug Sep Oct

Nov

Dec

1 0 21 7 5 7 6 0 1 0 1 2 1 2 2 1 0 1 1 0 9 1 2 1

37 57 43 19 10 9 13 12 16 14 12 9 8 18 3 11 12 18 12 14 0 15 29 13

8 20 6 17 21 24 12 22 22 26 24 0 13 22 5 21 27 20 16 20 1 20 9 21

103 6 0 0 13 0 0 12 5 0 0 0 0 0 0 0 0 3 0 0 0 0 5 0

0 15 69 80 33 24 34 32 56 39 59 56 64 59 16 80 68 48 81 64 0 79 77 38

16 5 29 6 19 36 34 19 15 20 10 0 25 15 16 18 26 25 0 17 35 7 0 0

11 0 46 15 31 22 15 16 8 13 1 1 11 10 4 16 17 14 0 14 23 10 0 0

0 18 17 0 0 7 4 4 1 2 2 0 12 6 0 3 0 3 6 3 0 4 0 0

14 0 10 11 13 1 14 8 9 6 6 18 14 16 2 8 11 14 0 0 3 5 7 11

54 0 37 29 30 39 27 20 25 37 62 30 33 31 35 35 30 30 24 4 14 42 28 25

9 17 9 6 2 0 27 5 5 16 3 14 4 15 9 5 3 7 6 8 0 7 18 10

10 7 3 4 1 3 0 1 4 2 6 0 0 6 0 3 9 7 4 3 0 3 9 3

55 19 0 0 12 0 0 9 4 0 0 0 0 0 0 0 0 3 8 0 0 1 7 0

0 0 47 42 14 11 15 14 27 24 32 30 39 29 11 48 47 36 46 39 0 46 50 33

4 0 7 1 1 17 15 6 8 10 10 0 9 10 2 6 13 12 1 10 31 1 0 0

1 0 19 2 5 1 4 2 0 1 0 0 12 9 0 8 7 7 5 9 8 3 0 0

17 0 0 0 0 0 0 0 0 0 0 0 39 0 0 0 0 0 0 0 0 0 0 0

No

No

No

No

Yes

No

No

No

No

Yes

No

No

No

Yes

No

No

No

ESu ISW > 89° W

Su IS

Fall ISW > 89°W

(61 diameter, 333 µm mesh size) towed in a doubleoblique manner to 200 m depth, or within 2 to 5 m of the bottom, as described in Richards et al. (1993) and Scott et al. (1993). Samples from only one bongo net were sorted, and the larvae were identified to the lowest possible taxa at the Sea Fisheries Institute Plankton Sorting and Identification Center, Gdynia and Szczecin, Poland. Catches of larvae were standardized to account for the volume filtered and depth of the sampled water column and were expressed as the number of larvae under 10 m2 of sea surface (Lyczkowski-Shultz & Hanisko 2007). Although > 500 taxa of larval fish were collected, species-level identifications were inconsistent through time. The reasons for this included limitations on larval fish identifications in the taxonomically rich Gulf of Mexico and improvements in the taxonomic

Sp OS

Su OS 64– 400 m

descriptions of some families since the surveys commenced in the early 1980s. All of the taxa were therefore aggregated to the family level, and the families with consistent identifications through time (M. Konieczna, Sea Fisheries Institute Plankton Sorting and Identification Center, Poland, pers. comm.) that contributed at least 0.5% of the total abundance, were selected for analysis (Table 2). These 20 families included 81.4% of all larvae collected throughout the time series analyzed. Identifications were reliable at the family level from 1984 onwards, with the exception of 2001, giving 24 yr of data in total. Although deeper waters were sampled during some spring surveys, only samples collected in less than 2000 m of depth were considered in this study. Environmental data were obtained from both in situ and remotely sensed datasets. An annual index

40


Table 2. The 20 larval fish families included in assemblage analyses, including percent contributions to total concentrations (no of ind. 10 m–2). Results of family-specific neural network models are also shown, with model R2, along with the relative importance of the 3 explanatory variables

blages and to highlight families with particularly strong trends (Fig. 2). However, an important initial problem was the temporal and spatial variability of sampling effort among Family Variable importance (out of 100) years (Fig. 1). Larval fish assemblages % Model Water Year Longitude may be expected to show strong contribution R2 depth (m) day seasonal and depth-related structure Myctophidae 14.65 0.762 100 3.4 1.6 (Grothues & Cowen 1999), and Bregmacerotidae 9.98 0.541 100 3.4 11.3 interannual inconsistencies in samGobiidae 9.29 0.431 100 27 21.2 pling thus have the potential to Engraulidae 8.31 0.462 49.4 87.1 100 Clupeidae 5.89 0.206 92.6 100 56.7 strongly confound underlying trends. Paralichthyidae 4.22 0.338 100 36.3 1.8 Attempts to address this by grouping Sciaenidae 4.06 0.529 100 72.6 49.5 samples into discrete categories in Synodontidae 3.91 0.427 100 11.7 3.4 time or space are often only partially Carangidae 3.88 0.305 21 100 2 Gonostomatidae 3.73 0.695 100 0.5 0.1 satisfactory, so we instead modeled Cynoglossidae 3.08 0.375 72.8 55.2 100 the influence of water depth, day of Bothidae 1.69 0.21 100 12.3 12.7 the year and longitude on the abunScombridae 1.68 0.338 74.9 100 20.4 dances of each family using multiSternoptychidae 1.68 0.481 100 0.5 0.2 layer perceptron neural network Ophidiidae 1.61 0.291 100 6.9 3 Phosichthyidae 0.90 0.501 100 4.6 0.5 models (Table 2), built in DTREG softLabridae 0.88 0.3325 33.3 38.9 100 ware (Sherrod 2003). These models Paralepididae 0.67 0.423 100 3.9 3.7 were well suited to our purpose beOphichthidae 0.66 0.22 100 57.2 54.8 cause they can work effectively with Scorpaenidae 0.59 0.324 97.5 59.4 100 the non-normal distributions typical of larval fish data (Segurado & Araujo of water temperature in the northern Gulf of Mexico 2004). To calculate the relative importance of each (25 to 30°N) was obtained from the HadISST product predictor variable, the improvement in classification (British Atmospheric Data Center). A 3 yr moving gained by each split that used the predictor was mean of temperature was also included. Mean summed, and the results were standardized to a monthly Mississippi River outflow was obtained from score out of 100 (Sherrod 2003). The observed larval the US Army Corps of Engineers (www.mvn.usace. concentrations were then subtracted from predicted army.mil/cgi-bin/watercontrol.pl?01160, accessed values to give residuals, which were used in all fur15 May 2011), and the annual spatial extent of the ther analyses (Fig. 3). These residuals thus described continental shelf hypoxic zone was obtained from departures from an expected long-term mean, given N. Rabalias (Louisiana Universities Marine Consorthe position in the Gulf of Mexico and day of the year. tium, LUMCON, pers. comm.). Annual shrimp trawlThe same procedure was then applied to plankton ing effort in days per year on the Gulf of Mexico shelf volume data because these were also strongly influwas sourced from NOAA-NMFS (Nance 2004, enced by water depth and, to a lesser extent, time of J. Nance pers. comm.). Lastly, total displaced plankyear. ton volumes per sample were obtained where availTo visualize the typical spatiotemporal distribuable and standardized to volumes per 10 m2 of seations of larvae from each family, predicted larval conwater sampled. centrations (no. of ind. 10 m−2) were calculated for the fifteenth day of each month (April to December), at 20 water depths between 20 and 2000 m, using the Data analysis neural network models created for each family. All predicted concentrations were divided by the maxiObserved vs. predicted larval anomalies mum values within each family, giving a maximum predicted concentration of 1, to allow comparison of Larval concentrations were log10(x + 1) transmore abundant and less abundant families. These formed prior to analysis, to address strong rightstandardized predicted larval concentrations were skewed distribution. These data were then used to then kriged in Surfer 9 (Golden Software) using day describe decadal-scale changes in larval fish assemof the year as the x-axis and water depth as the


41

Log10-transformed larval fish data from surveys (Fig. 1, Table 1)

Neural network models to remove spatiotemporal bias (Table 2, Fig. 3)

Residual larval concentrations (observed/predicted from neural network models)

Do larval fish assemblages change through time? Test: multivariate index of seriation (Table 3)

How do peak spawning times and areas differ among families? Test: predictions from neural network models (Figs. 4 & 5)

How do patterns differ for inner and outer shelf samples? (Fig. 10)

Which families change the most strongly? Test: Pearson correlation with year (Table 3, Fig. 6) Which families vary in synchrony? Test: cluster analysis (Fig. 7)

Is this change correlated with environmental conditions? Test: DISTLM (Figs. 8 & 9)

Fig. 2. Data sources (boxes), analyses completed and study objectives

y-axis, to allow the comparison of spatiotemporal peaks in larval abundance among families.

Examining change through time Because shallow- and deep-water families were likely to have different responses to environmental variability, predicted depth-abundance curves were examined to determine an appropriate depth to separate ‘inner shelf’ and ‘outer shelf’ zones. The historical sampling coverage was then examined for each month and depth zone. Months with at least 10 samples collected in at least 20 yr of the survey were selected to use as discreet time series, with years of low coverage ( 60% of the interannual environmental variability, with axis 1 describing the directional increase in surface temperatures and the decrease in trawling effort and plankton volumes and axis 2 describing the interannual variability from Mississippi River outflow and the extent of the hypoxic zone. Patterns of environmental variability explained modest but significant amounts of variability in the larval fish assemblages (Fig. 9). Inner-shelf assemblages were best correlated with a combination of annual trawling effort (p = 0.012), Mississippi River outflow (p = 0.06) (both detrended) and a variable describing a simple linear trend (p = 0.001). Overall, this combination explained 15.7%


Table 3. Multivariate index of seriation analysis, for each of 5 time series. ρ statistics and p-values are shown, along with Pearson correlation coefficients (highest only) between mean residual family larval abundances and year ρ

p

Family

Corr. coeff.

Inner shelf, (west) early summer

0.248

0.007

Bothidae Cynoglossidae Sciaenidae

−0.748 −0.515 0.489

Inner shelf, (west) fall

0.290

0.006

Clupeidae Paralichthyidae Sciaenidae

0.557 0.622 0.642

Inner shelf, summer

0.248

0.008

Bothidae Cynoglossidae Sciaenidae

−0.581 −0.587 0.798

Outer shelf, spring

0.343

0.002

Gonostomatidae Labridae Scombridae

0.618 0.621 0.664

Outer shelf, summer

0.269

0.002

Gonostomatidae Phosichthyidae Scombridae

0.765 0.661 0.753

Time series

45

of the observed variability in larval fish assemblages. Concentrations of Bothidae, Cynoglossidae and Sciaenidae larvae were influential in driving the obser ved patterns, with Bothidae and Cynoglossidae larvae tending to decrease in abundance between the 1980s and 2000s (as shown by the inverse relationship with the positive linear trend), while Sciaenidae larvae increased (Fig. 9). Outer-shelf assemblages were best correlated with a combination of a 3 yr moving mean of sea surface temperature (p = 0.001), outer-shelf plankton volumes (p = 0.11) (both detrended) and a variable describing a simple linear trend (p = 0.001),

Fig. 6. Pearson correlation coefficients of residual larval abundances (no. of ind. 10 m−2) at rounded 1° × 1° stations, kriged across the northern Gulf of Mexico. Six families highlighted in the multivariate index of seriation analyses are shown. Blue areas indicate a decrease over the course of the survey period (1984 to 2008), while red and yellow areas indicate an increase. Examples of trends in residual concentrations of Bothidae and Gonostomatidae larvae through time at specific grid points are shown


46

Correlation

–0.5

Group Benthic Pelagic Inshore Mesopelagic

0

0.5

Carangidae

0.4

0.6

Summer inner shelf

0.4

Scombridae

Synodontidae

Bregmacerotidae

Ophidiidae

Gonostomatidae

Paralichthyidae

Ophichthidae

Cynoglossidae

Bothidae

Gobiidae

Myctophidae

Gonostomatidae

Paralepididae

Phosichthyidae

Scorpaenidae

Scombridae

Sternoptychidae

Labridae

Carangidae

Engraulidae

Clupeidae

Mean residual concentration (no. of ind. 10 m–2)

0.6

Sciaenidae

1.0

Summer outer shelf

Myctophidae

0.2 0.2

0 –0.2

0

–0.4 –0.2 –0.6 –0.4 1980

1985

1990

1995

2000

2005

2010

–0.8 1980

1985

1990

1995

2000

2005

2010

Year Fig. 7. Cluster analysis of larval fish families computed on a resemblance matrix of Pearson correlation coefficients of residual larval abundances, averaged for time series and year. Examples of covarying families are also shown for 2 inner-shelf and 2 outer-shelf families

which together explained 27.1% of the observed variability. The concentrations of the mesopelagic families Gonostomatidae, Myctophidae and Phosichthyidae were influential in driving the patterns observed, as were Cynoglossidae and Scombridae larvae (Fig. 9). All of these families were positively correlated with the variable describing a linear trend, suggesting a general increase in abundance through time. Concentrations of both Carangidae and Scombridae larvae were correlated to a moving mean of the sea surface temperature, and all tended to increase between the 1980s and 2000s. In contrast, the increases observed in the concentrations of mesopelagic families were not strongly associated with any of the environmental predictors considered.

To summarize changes in assemblage structure through time for both the inner and outer continental shelf, multivariate indices were developed by averaging the residuals of family abundance for each time series, in each year (Fig. 10). Inner-shelf assemblages showed some changes between samples from the 1980s, 1990s and 2000s; however, outer-shelf assemblages showed a stronger directional change through time (Fig. 10).

DISCUSSION Larval fish assemblages in the northern Gulf of Mexico were strongly structured by water depth and

Plankton: Plankton: inner shelf outer shelf

Annual river outflow

Plankton: inner shelf

b

North GOM Hypoxia SST moving SST mean 26.0

Trawling effort

390 000 340 000

0.2

290 000

0.0 –0.2

240 000

–0.4

190 000

–0.6

140 000

–0.8

90 000

–1.0

40 000

SST moving mean Hypoxic zone

c

25.8 25.6

30 000

25.4

25 000

25.2

20 000

25.0

15 000

24.8 10 000

24.6 24.4

5000

24.2

0

1980 1985 1990 1995 2000 2005 2010

1980 1985 1990 1995 2000 2005 2010

Year

PCO2 (22.3% of total variation)

5

Year

d Hypoxic zone

Decade 1980s 1990s 2000s

River outflow

Plankton: inner shelf Plankton: outer shelf

Surface temperature 0

Shrimp-trawling effort SST moving mean

–5 –6

35 000

Area of hypoxoc zone (km2)

Trawling

0.4

Plankton: inner shelf (residual ml m–3)

47

a

Northern GOM SST (°C)

0.6

1 0.8 0.6 0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1

Shrimp-trawl fishing effort (d)

Pearson correlation coeficent


–4

–2

0

2

4


season. This pattern is consistent with findings from other regions of the world (e.g. Doyle et al. 1993, Hare et al. 2001, Duffy-Anderson et al. 2006) and usually reflects the depth preferences and spawning seasons of adult fishes. Although migratory movements of adults may complicate matters (e.g. Ophichthidae,

6

Fig. 8. (a) Pearson correlation coefficients of environmental variables regressed against year. Significant changes (p < 0.05, corrected for multiple comparisons) are shown with black bars. Trends through time of (b) plankton abundance (ml plankton per m3 seawater sampled) and shrimp trawling effort and (c) the northern Gulf of Mexico (GOM) sea surface temperature (SST, 3 yr moving average) and hypoxic zone size from 1984 to 2008. (d) Principal coordinates (PCO) analysis ordination of mean annual environmental conditions in the northern Gulf of Mexico, 1984 to 2008

which move offshore to spawn) (Ross & Rohde 2003), and larvae are inevitably mixed and dispersed by ocean currents (Grothues & Cowen 1999, Muhling et al. 2008), the spatial and temporal characteristics of larval fish assemblages are usually strongly apparent. The oceanography of the northern Gulf of Mexico


48

Inner shelf a

Trawling effort (detrended)

dbRDA2 (23.3% of fitted, 6.3% of total variation)

River outflow (detrended)

0.5

0 Linear trend

–0.5

Plankton (detrended) 1

0

–1

–1.5

SST moving mean (detrended)

–1.0

–0.5

0

0.5

1.0

–3

1.5

2



1.0

c Cynoglossidae

0.5 Sciaenidae 0

Bothidae

–0.5

–2

–1

0

1

2

3



d

1

Cynoglossidae

Myctophidae 0 Phosichthyidae Gonostomatidae –1

Scombridae

–2

–1.0 –1.5

–1.0

– 0.5

0

0.5

1.0

–3

1.5

300 250

0.2

200

0 –0.2

150

–0.4

100 50

–0.6 –0.8 1980

1985

1990

1995

Year

2000

2005

0 2010

Mean residual larval concentration (no. of ind. 10 m–2)

0.4

350

Northern GOM shrimp trawling (1000 d)

Bothidae: inner shelf, early summer Annual trawling effort

e

–2

–1

0

1

2

3


dbRDA1 (60.6% of fitted, 9.3% of total variation) Mean residual larval concentration (no. of ind. 10 m–2)

Linear trend

–2

–1.0

0.6

b

0.6

26

f

26 0.4

26

0.2

25 25

0 Scombridae: outer shelf, spring SST mean

–0.2 –0.4 1980

1985

1990

1995

2000

2005

25 25

25 2010

Year

Fig. 9. Results of distance-based linear models of residual larval fish assemblages for (a,c,e) inner-shelf and (b,d,f) outer-shelf zones, displayed using distance-based redundancy analysis. Axes describe the percentage of variation described in terms of the selected model and in terms of the total variation of the larval fish assemblage. (a,b) Vectors representing the environmental variables comprising the best selected models (from Akaike’s information criterion). (c,d) Vectors representing families with high (> 0.5) correlation with axes. (e,f) Examples of correlations highlighted by the multivariate analysis, with (e) residual abundances of Bothidae larvae against shrimp trawling effort and (f) residual abundances of Scombridae larvae against a 3 yr running mean of sea surface temperature (SST), for 2 selected time series

Northern GOM SST: 3 yr mean (°C)


1.0

Outer shelf 2


Inner shelf 1.0

1986

2006

2000 0.5

1988

1999

1991

2002 0 1993 –0.5

1992 1997 1990

–1.0 –2.0

2004



Outer shelf

1990

2

49

2006

20032007

1

19992008

1986

1993

1987 0

2000 2002 1994

Decade 1980s 1990s 2000s

–1

1988

1996 19971995 1989

1998

–2 –1.5

–1.0

–0.5

0

0.5


1.0

–4

–3

–2

–1

0

1


Fig. 10. Multivariate annual indicators derived from residual larval fish abundances in inshore (left) and offshore (right) waters, displayed using principal coordinates (PCOs) analysis

generally favors retention of larvae over the continental shelf, for time scales relevant to typical pelagic larval durations (Ohlmann & Niiler 2005). This retention is likely to be responsible for the strong crossshelf gradient that structured assemblages. Once these signals had been largely removed from the dataset, interannual and decadal-scale changes became apparent. Our original hypothesis that larvae of benthic families would decrease was partially supported, but only for some families. In addition, larvae of some other inner-shelf families increased strongly. Also noticeable was the tendency for larvae of mesopelagic families to increase between the start and end of the survey period as well as the decrease in abundance of 2 common inshore families (Bothidae and Cynoglossidae), which partially supported the hypothesis that larvae of fish families occupying similar habitats as adults would show similar decadal-scale changes across the time series. Although correlations between larval fish assemblages and environmental variables suggested that the drivers behind patterns observed are likely to be complex and are not fully defined in this study, some candidate processes for further examination emerged. The effect of shrimp trawling effort was the strongest and most significant predictor of interannual variability in inner-shelf larval fish assemblages. Trawling effort on the northern Gulf of Mexico shelf has declined since the 1980s, particularly in the 2000s. Trawling was estimated to be a significant source of mortality for sciaenid populations (Diamond et al. 2000), and the decline in trawling effort may have contributed to the observed increase in sci-

aenid populations (as estimated using larval abundances). However, this reduction in effort has clearly not had a similarly positive impact on Bothidae and Cynoglossidae populations. Rijnsdorp & Vingerhoed (2001) showed that trawling in the North Sea may actually improve feeding conditions for some flatfish, which tend to target short-lived, productive benthic organisms. Benthic fish species with different dietary preferences may thus have inverse responses to bottom disturbance from trawling effort, and this may be the case here. In addition, habitat condition for some adult benthic fishes may have been influential, as the decrease in most benthic families was most pronounced in the region of the summer hypoxic zone (Rabalais et al. 2002). Flatfishes generally avoid hypoxic areas, and their main food sources may also be affected by hypoxic events (Switzer et al. 2009). However, in contrast, Sciaenidae larvae increased in abundance in inshore waters. Because sciaenids are also sensitive to hypoxic events (Craig & Crowder 2005), it appears more likely from our data that disturbance from shrimp trawling was the dominant driver of change for the benthic fish communities. In addition, outflow from the Mississippi River explained some of the interannual variability in inner-shelf assemblage structure, likely as a consequence of the freshwater plume’s influence on primary production (Lohrenz et al. 1997, Dagg & Breed 2003). Drivers behind the observed increases in abundances of mesopelagic larvae were more difficult to elucidate. Although increases in abundance of mesopelagic families in concert with environmental

2

50


changes have been recorded previously (Smith & Moser 2003, Watanabe & Kawaguchi 2003), interannual variability of tropical North Atlantic and Gulf of Mexico temperatures were not well correlated to observed increases in mesopelagic larvae (Wang & Enfield 2001). Conley & Hopkins (2004) suggested that Myctophidae larvae, the most abundant in our collections, were not likely to be prey-limited in the eastern Gulf of Mexico, and although detailed zooplankton assemblage data were not available to us, the total volumes of plankton from SEAMAP samples have actually decreased somewhat since the early 1980s. We therefore did not find evidence of largescale changes in feeding conditions in the Gulf of Mexico, although we did not have the data necessary to evaluate this possibility in any depth. Mesopelagic fishes are highly abundant in the Gulf of Mexico, as well as in many other world oceans (Sassa et al. 2002, Doyle et al. 1993, Moser & Smith 1993), and thus represent an important food source for higher predators. They are consumed by other fishes, cephalopods, marine birds and mammals (Cherel et al. 1993, Das et al. 2000, Kitchell et al. 2002, Hinke et al. 2004), with the relative importance of these predators varying by region. It is therefore possible that reduced predation on either larval or adult stages of mesopelagic fishes led to the observed increases. Variability in predation pressures on either larval or adult stages is very difficult to quantify; however, we note that mesopelagic fishes may form a major component of the food webs supporting juvenile and adult tunas and swordfishes in some regions (Menard et al. 2000, Moteki et al. 2001, Hinke et al. 2004). In addition, changes in predator populations may significantly impact abundances of mesopelagic prey species (Essington et al. 2002). Indicators of abundance for these predators in the Gulf of Mexico are difficult to obtain for the length of the larval time series; however, spawning stock biomass estimates for yellowfin tuna Thunnus albacares, bigeye tuna T. obesus and swordfish Xiphias gladius in the western Atlantic Ocean have all declined since the mid-1980s (ICCAT 2009, 2010, 2011). These species all inhabit the Gulf of Mexico, and all have been found to rely heavily on mesopelagic fishes as part of their diets (Moteki et al. 2001), through linkages with cephalopods and smaller scombrid fishes (Hinke et al. 2004). Varying rates of predation on the larvae of mesopelagic species may also be influential but are extremely difficult to assess and are beyond the scope of this study. These connections should be investigated in more depth in future studies, within the context of Gulf of Mexico food webs.

Abundances of the larvae of some smaller pelagic families (Carangidae and Scombridae) also increased through time. While both families are speciose, the majority of carangid larvae were likely to have been Chloroscombrus chrysurus or Decapterus punctatus (Ditty et al. 2004), while the scombrid larvae identified to genus were most commonly Scomberomorus, Auxis, Euthynnus and Thunnus species. Larvae of blackfin (Thunnus atlanticus) and yellowfin (Thunnus albacares) tuna are very difficult to identify visually; however, most Thunnus larvae in the SEAMAP samples were likely blackfin tuna (W. Richards, NOAA-NMFS, pers. comm.). Seasonal abundances of both Carangidae and Scombridae larvae suggested that adults were predominantly summer spawners, with Carangidae larvae more abundant inshore and Scombridae larvae more abundant offshore. Interannual variability in the abundances of both families was correlated to anomalies in sea surface temperature, with warmer years resulting in higher abundances of larvae. This may have been a result of an extended spawning season in warmer years, as water temperatures have been linked to the onset of spawning activity (McPherson 1991, Itano 2000, Muhling et al. 2010) and larval growth rates (Sanchez-Ramirez & Flores-Coto 1998). It is also possible that reduced predation pressure on smaller, less exploited scombrid fishes (e.g. T. atlanticus, Auxis spp., etc.) by larger, exploited species (e.g. T. albacares and Xiphias gladius) may be significant, as described above. However, given the current lack of detailed information on Gulf of Mexico food webs, these connections must necessarily remain speculative. As anthropogenic pressures continue to exert influence on the Gulf of Mexico ecosystem, it appears likely that in the short to medium term, temperatures will continue to increase, the size of the hypoxic zone will continue to grow, and trawling activity will remain at historically low levels (Rabalais et al. 2002, Hansen et al. 2006, Caillouet et al. 2008). Results from this study suggest that summer-spawning species, tolerant of hypoxia or able to move away from hypoxic areas, which are not reliant on bottom disturbance, may be favored. However, species-specific studies are needed to examine these issues in more detail because within families, or even genera, responses of species to environmental conditions can be markedly different. As an example, scombrid fishes in the Thunnus genus may include the highly migratory bluefin tuna Thunnus thynnus, which is intolerant of warmer waters and spawns in the Gulf of Mexico


in spring (Teo et al. 2007), and the more inshoreliving blackfin tuna T. atlanticus, which is a tropical species that spawns in summer. Analyses at the family level, such as the one presented here, thus provide only a starting point for more in-depth, species-level examinations. The Gulf of Mexico ecosystem is therefore driven by a complex suite of natural and anthropogenic drivers, which impact habitat suitability, predation pressure and entire food webs. Extractive fisheries selectively remove valuable species and impact habitat quality, while nutrient inputs from riverine sources increase productivity and improve feeding conditions for some species, while causing hypoxia, toxic algal blooms and habitat loss for others (Rabalais et al. 2002, Dagg & Breed 2003, Heil et al. 2007). Climate change may be beginning to result in warming temperatures, with potential for changes in the assemblages of temperature-sensitive species (Cuevas et al. 2004, Fodrie et al. 2010) and the growth and survival of larvae (Wilderbuer et al. 2002). In spring 2010, the Deepwater Horizon oil spill released large amounts of oil into the Gulf of Mexico ecosystem, the biological effects of which are still unclear and may remain so for decades. The need for easily collectable and interpretable indicators is therefore significant and immediate. Larval fish surveys are more easily completed and costeffective than comprehensive surveys of adult fishes across both benthic and pelagic habitats. The larval fish assemblage indicators developed here thus encompass a wide variety of fish functional groups, a valuable property given the range of responses of fish taxa with differing life histories to environmental forcing (Auth et al. 2011), and provide a starting point for future studies of species-specific responses and food web structure.

Acknowledgements. The authors acknowledge the support and assistance of staff at NOAA-NMFS, especially J. Lyczkowski-Shultz, D. Hanisko, G. Zapfe, D. Drass and P. Bond, and K. Williams from the Fish and Wildlife Research Institute, for providing data, helpful information and suggestions. We also thank M. Konieczna and L. Ejsymont from the Sea Fisheries Institute Plankton Sorting and Identification Center, Gdynia and Szczecin, Poland, and the captains and crew of all the NOAA ships who collected data on the SEAMAP cruises. G. Pellegrin and J. Nance provided fisheries and trawl survey data, while N. Rabalais provided data on the extent of the hypoxic zone. Sea surface temperature data were obtained with the assistance of S-K. Lee and Y. Liu from NOAA-AOML, and D. Enfield provided valuable assistance and advice on analyses. We also thank the 5 anonymous reviewers, whose comments improved the manuscript.

51

LITERATURE CITED

➤

➤

➤ ➤

➤

➤

➤

➤

➤ ➤

Alverson DL, Freeberg MH, Murawski SA, Pope JG (1996) A global assessment of fisheries bycatch and discards. FAO Fish Tech Pap 339. FAO, Rome Anderson MJ (2001) A new method for non-parametric multivariate analysis of variance. Austral Ecol 26:32−46 Anderson PJ, Piatt JF (1999) Community reorganization in the Gulf of Alaska following ocean climate regime shift. Mar Ecol Prog Ser 189:117−123 ICCAT (International Commission for the Conservation of Atlantic Tunas) (2009) Report of the 2008 ICCAT yellowfin and skipjack stock assessments meeting. Collect Vol Sci Pap ICCAT 64:669−927 ICCAT (2010) Report of the 2009 Atlantic swordfish stock assessment session. Collect Vol Sci Pap ICCAT 65: 1−123 ICCAT (2011) Report of the 2010 ICCAT bigeye tuna stock assessment session. Collect Vol Sci Pap ICCAT 66:1−186 Auth TD, Brodeur RD, Soulen HL, Ciannelli L, Peterson WT (2011) The response of fish larvae to decadal changes in environmental forcing factors off the Oregon coast. Fish Oceanogr 20:314−328 Beaugrand G (2004) The North Sea regime shift: evidence, causes, mechanisms and consequences. Prog Oceanogr 60:245−262 Caillouet CW Jr, Hart RA, Nance JM (2008) Growth overfishing in the brown shrimp fishery of Texas, Louisiana, and adjoining Gulf of Mexico EEZ. Fish Res 92:289−302 Cherel Y, Verdon C, Ridoux V (1993) Seasonal importance of oceanic myctophids in king penguin diet at Crozet Islands. Polar Biol 13:355−357 Clarke KR (1993) Non-parametric multivariate analyses of changes in community structure. Aust J Ecol 18:117−143 Clarke KR, Gorley RN (2006) PRIMER v6: User manual/tutorial. PRIMER-E, Plymouth Clarke KR, Warwick RM, Brown BE (1993) An index showing breakdown of seriation, related to disturbance, in a coral-reef assemblage. Mar Ecol Prog Ser 102:153−160 Conley WJ, Hopkins TL (2004) Feeding ecology of lanternfish (Pisces: Myctophidae) larvae: prey preferences as a reflection of morphology. Bull Mar Sci 75:361−379 Craig JK, Crowder LB (2005) Hypoxia-induced habitat shifts and energetic consequences in Atlantic croaker and brown shrimp on the Gulf of Mexico shelf. Mar Ecol Prog Ser 294:79−94 Cuevas KJ, Franks JS, Buchanan MV (2004) First record of bone-fish, Albula vulpes, from Mississippi coastal waters. Gulf Caribb Res 17:69−94 Dagg MJ, Breed GA (2003) Biological effects of Mississippi River nitrogen on the northern Gulf of Mexico — a review and synthesis. J Mar Syst 43:133−152 Das K, Lepoint G, Loizeau V, Debacker V, Dauby P, Bouquegneau JM (2000) Tuna and dolphin associations in the North-east Atlantic: evidence of different ecological niches from stable isotope and heavy metal measurements. Mar Pollut Bull 40:102−109 Diamond SL, Cowell LG, Crowder LB (2000) Population effects of shrimp trawl bycatch on Atlantic croaker. Can J Fish Aquat Sci 57:2010−2021 Ditty JG, Shaw RF, Cope JS (2004) Distribution of carangid larvae (Teleostei: Carangidae) and concentrations of zooplankton in the northern Gulf of Mexico, with illustrations of early Hemicaranx amblyrhynchus and Caranx spp. larvae. Mar Biol 145:1001−1014

52

➤

➤

➤

➤ ➤ ➤ ➤

➤

➤

➤

➤ ➤


Doyle MJ, Morse WW, Kendall AW Jr (1993) A comparison of larval fish assemblages in the temperate zone of the northeast Pacific and northwest Atlantic Oceans. Bull Mar Sci 53:588−644 Duffy-Anderson JT, Busby MS, Mier KL, Deliyanides CM, Stabeno PJ (2006) Spatial and temporal patterns in summer ichthyoplankton assemblages on the eastern Bering Sea shelf 1996−2000. Fish Oceanogr 15:80−94 Essington TE, Schindler DE, Olson RJ, Kitchell JF, Boggs C, Hilborn R (2002) Alternative fisheries and the predation rate of yellowfin tuna in the eastern Pacific Ocean. Ecol Appl 12:724−734 Fodrie FJ, Heck KL Jr, Powers SP, Graham WM, Robinson KL (2010) Climate-related, decadal-scale assemblage changes in seagrass-associated fishes in the northern Gulf of Mexico. Glob Change Biol 16:48−59 Francis RC, Hare SR (1994) Decadal-scale regime shifts in the large marine ecosystems of the north Pacific: a case for historical science. Fish Oceanogr 3:279−291 Grothues TM, Cowen RK (1999) Larval fish assemblages and water mass history in a major faunal transition zone. Cont Shelf Res 19:1171−1198 Hansen J, Sato M, Ruedy R, Lo K, Lea DW, Medina-Elizade M (2006) Global temperature change. Proc Natl Acad Sci USA 103:14288−14293 Hare JA, Fahay MP, Cowen R (2001) Springtime ichthyoplankton of the slope region off the north-eastern United States of America: larval assemblages, relation to hydrography and implications for larval transport. Fish Oceanogr 10:164−192 Heil CA, Revill M, Glibert PM, Murasko S (2007) Nutrient quality drives differential phytoplankton community composition on the southwest Florida shelf. Limnol Oceanogr 52:1067−1078 Hilty J, Merenlender A (2000) Faunal indicator taxa selection for monitoring ecosystem health. Biol Conserv 92: 185−197 Hinke JT, Kaplan IC, Aydin K, Watters GM, Olson RJ, Kitchell JF (2004) Visualizing the food-web effects of fishing for tunas in the Pacific Ocean. Ecol Soc 9:10, available at www.ecologyandsociety.org/vol9/iss1/art10/ Itano DG (2000) The reproductive biology of yellowfin tuna (Thunnus albacares) in Hawaiian waters and the western tropical Pacific Ocean: project summary. SOEST 00-01 JIMAR Contribution 00-328. Joint Institute for Marine and Atmospheric Research, Honolulu, HI Kitchell JF, Essington TE, Boggs CH, Schindler DE, Walters CJ (2002) The role of sharks and longline fisheries in a pelagic ecosystem of the central Pacific. Ecosystems 5: 202−216 Levin PS, Holmes EE, Piner KR, Harvey CJ (2006) Shifts in a Pacific Ocean fish assemblage: the potential influence of exploitation. Conserv Biol 20:1181−1190 Lohrenz SE, Fahnenstiel GL, Redalje DG, Lang GA, Chen X, Dagg MJ (1997) Variations in primary production of northern Gulf of Mexico continental shelf waters linked to nutrient inputs from the Mississippi River. Mar Ecol Prog Ser 155:45−54 Lyczkowski-Shultz J, Hanisko DS (2007) A time series of observations on Red Snapper larvae from SEAMAP surveys, 1982-2003: seasonal occurrence, distribution, abundance and size. In: Patterson WF III, Cowan JH Jr, Fitzhugh GR, Nieland DL (eds) Red snapper ecology and fisheries in the U.S. Gulf of Mexico. Am Fish Soc Symp 60:3−23

➤ McPherson GR (1991) Reproductive biology of yellowfin

➤

➤

➤

➤

➤

➤ ➤

➤

➤

tuna in the eastern Australian fishing zone, with special reference to the north-western Coral Sea. Mar Freshw Res 42:465−477 Menard F, Stequert B, Rubin A, Herrere M, Marchal E (2000) Food consumption of tuna in the Equatorial Atlantic ocean: FAD-associated versus unassociated schools. Aquat Living Resour 13:233−240 Moser HG, Smith PE (1993) Larval fish assemblages of the California Current region and their horizontal and vertical distributions across a front. Bull Mar Sci 53: 645−691 Moteki M, Arai M, Tsuchiya K, Okamoto H (2001) Composition of piscine prey in the diet of large pelagic fish in the eastern tropical Pacific Ocean. Fish Sci 67:1060−1074 Muhling BA, Beckley LE, Koslow JA, Pearce AF (2008) Larval fish assemblages and water mass structure off the oligotrophic south-western Australian coast. Fish Oceanogr 17:16−31 Muhling BA, Lamkin JT, Roffer MA (2010) Predicting the occurrence of Atlantic bluefin tuna (Thunnus thynnus) larvae in the northern Gulf of Mexico: building a classification model from archival data. Fish Oceanogr 19: 526−539 Müller-Karger FE, Walsh JJ, Evans RH, Meyers MB (1991) On the seasonal phytoplankton concentration and sea surface temperature cycles of the Gulf of Mexico as determined by satellites. J Geophys Res C 96: 12645−12665 Nance J (2004) Estimation of effort in the offshore shrimp trawl fishery of the Gulf of Mexico. Red Snapper SEDAR Data Workshop, April 2004, SEDAR7-DW-24, Charleston, SC Nye JA, Link JS, Hare JA, Overholtz WJ (2009) Changing spatial distribution of fish stocks in relation to climate and population size on the Northeast United States continental shelf. Mar Ecol Prog Ser 393:111−129 Ohlmann JC, Niiler PP (2005) Circulation over the continental shelf in the northern Gulf of Mexico. Prog Oceanogr 64:45−81 Rabalais NN, Turner RE, Wiseman WJ Jr (2002) Gulf of Mexico hypoxia, A.K.A. ‘The Dead Zone. Annu Rev Ecol Syst 33:235−263 Richards WJ, McGowan MF, Leming T, Lamkin JT, Kelley S (1993) Larval fish assemblages at the Loop Current boundary in the Gulf of Mexico. Bull Mar Sci 53:475−537 Rijnsdorp AD, Vingerhoed B (2001) Feeding of plaice Pleuronectes platessa L. and sole Solea solea (L.) in relation to the effects of bottom trawling. J Sea Res 45:219−229 Ross SW, Rohde FC (2003) Collections of ophichthid eels on the surface at night off North Carolina. Bull Mar Sci 72: 241−246 Sanchez-Ramirez M, Flores-Coto C (1998) Growth and mortality of larval Atlantic Bumper Chloroscombrus chrysurus (Pisces: Carangidae) in the southern Gulf of Mexico. Bull Mar Sci 63:295−304 Sassa C, Moser HG, Kawaguchi K (2002) Horizontal and vertical distribution patterns of larval myctophid fishes in the Kuroshio Current region. Fish Oceanogr 11:1−10 Scott GP, Turner SC, Grimes CB, Richards WJ, Brothers EB (1993) Indices of larval bluefin tuna, Thunnus thynnus, abundance in the Gulf of Mexico; modeling variability in growth, mortality, and gear selectivity: ichthyoplankton methods for estimating fish biomass. Bull Mar Sci 53: 912−929


➤ Segurado P, Araujo MB (2004) An evaluation of methods

➤

➤

➤

for modelling species distributions. J Biogeogr 31: 1555−1568 Sherrod PH (2003) DTREG: Classification and regression trees for data mining and modeling. Available at www. dtreg.com/DTREG.pdf Smith PE, Moser HG (2003) Long-term trends and variability in the larvae of Pacific sardine and associated fish species of the California Current region. Deep-Sea Res II 50: 2519−2536 Switzer TS, Chesney EJ, Baltz DM (2009) Habitat selection by flatfishes in the northern Gulf of Mexico: implications for susceptibility to hypoxia. J Exp Mar Biol Ecol 381: S51−S64 Teo SLH, Boustany A, Dewar H, Stokesbury MJW and othEditorial responsibility: Kenneth Sherman, Narragansett, Rhode Island, USA

➤ ➤

➤

53

ers (2007) Annual migrations, diving behavior, and thermal biology of Atlantic bluefin tuna, Thunnus thynnus, on their Gulf of Mexico breeding grounds. Mar Biol 151: 1−18 Wang C, Enfield D (2001) The tropical Western hemisphere warm pool. Geophys Res Lett 28:1635−1638 Watanabe H, Kawaguchi K (2003) Decadal change in abundance of surface migratory myctophid fishes in the Kuroshio region from 1957 to 1994. Fish Oceanogr 12: 100−111 Wilderbuer TK, Hollowed AB, Ingraham WJ Jr, Spencer PD, Conners ME, Bond NA, Walters GE (2002) Flatfish recruitment response to decadal climatic variability and ocean conditions in the eastern Bering Sea. Prog Oceanogr 55:235−247 Submitted: July 1, 2011; Accepted: December 2, 2011 Proofs received from author(s): March 19, 2012